Abstract. General circulation models (GCMs) have been criticized for their failure to represent the observed scales of precipitation, particularly in the tropics where simulated daily rainfall is too light, too frequent and too persistent. Previous assessments have focused on temporally or spatially averaged precipitation, such as daily means or regional averages. These evaluations offer little actionable information for model developers, because the interactions between the resolved dynamics and parameterized physics that produce precipitation occur at the native gridscale and time step.We introduce a set of diagnostics ("Analyzing Scales of Precipitation", version 1.0 -ASoP1) to compare the spatial and temporal scales of precipitation across GCMs and observations, which can be applied to data ranging from the gridscale and time step to regional and sub-monthly averages. ASoP1 measures the spectrum of precipitation intensity, temporal variability as a function of intensity and spatial and temporal coherence. When applied to time step, gridscale tropical precipitation from 10 GCMs, the diagnostics reveal that, far from the "dreary" persistent light rainfall implied by daily mean data, most models produce a broad range of time step intensities that span 1-100 mm day −1 . Models show widely varying spatial and temporal scales of time step precipitation. Several GCMs show concerning quasi-random behavior that may influence and/or alter the spectrum of atmospheric waves. Averaging precipitation to a common spatial (≈ 600 km) or temporal (3 h) resolution substantially reduces variability among models, demonstrating that averaging hides a wealth of information about intrinsic model behavior. When compared against satellite-derived analyses at these scales, all models produce features that are too large and too persistent.